LOGO
INCYSPEAK
Semantic memory for machines.

Machines that learn from experience.

IncySpeak is a semantic-memory runtime for machines. An IncySpeak machine takes in experience, grounds it into entities and predicates, stores it as episodic memory, accumulates evidence over time, and forms semantic beliefs that it can predict from, revise, and explain.

WATCH INCYSPEAK LEARN →


The learning pipeline

1. experienceRaw input observed from configured sources.
2. language / event perceptionText and events are parsed into structured signals.
3. entity / predicate groundingSignals are grounded to subjects, predicates, and objects.
4. episodic memoryGrounded observations are stored with time and provenance.
5. evidence accumulationRepeated support and counter-evidence are weighed.
6. semantic beliefsStable statements emerge with a confidence and stage.
7. prediction / revision / provenanceBeliefs drive predictions and can be revised and explained.

What this site is

This repository is only the cloud side of IncySpeak:

  • the public IncySpeak website;
  • the authenticated web control plane; and
  • the cloud coordination / API layer for remote IncySpeak machines.

The semantic engine itself is not part of this site. It runs separately on ARM64 machines, which keep ownership of their local episodic and semantic memory. See how it works and the developer notes.